Facing climate change and rapid urbanization, urban flooding has exposed human and properties to increasing disaster risks. The attention from researchers and decision‐makers to understand the key role of flood regulation service (FRS) in flood management has arisen. However, the mechanism of FRS supply–demand is little known from landscape scale. The FRS assessment methodology considering interacts between source, sink, and flow landscape was proposed in this study. The spatial distributions of surface runoff generation, runoff reduction capacity, and flood inundation were mapped using one‐dimensional rainfall–runoff method SCS‐CN and two‐dimensional flood propagation model CADDIES. Four 3‐hour designed rainfall scenarios ranging from nuisance to extreme events (3a, 11a, 56a, and 100a) were simulated. The Liuyang River Watershed in Changsha Municipality, China was selected for case study. The results showed that, the differences of runoff reduction coefficient and runoff generation volume between vegetation and built‐up landscape have shortened. The peak flood depth, extent of flood inundation, and peak flood velocity have increased continuously with the growing rainfall intensity. The number of source–sink mismatch catchment was the highest under 56 and 100a, and the most of source‐sink match catchments were observed under 3a. Under four rainfall scenarios, the changes of source–sink relationships were witnessed and the potentials of flow zone in source–sink mismatch catchments have increased. The FRS management framework concerning supply–demand connections has been proposed based on source–sink–flow analysis. These findings could provide a scientific basis for sustainable urban flood management and disaster risk mitigation.